Bug #11829: [PERF] Benchmark of vm1_yield

• Status: Closed
• Assignee: -
• Target version: -
• Backport: 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: DONTNEED

[tgxworld (Guo Xiang Tan)]

ruby 2.3.0dev (2015-10-10 trunk 52102)

total: 4 trial(s) (2 trial(s) for 2 benchmark(s))
....

benchmark results:
Execution time (sec)
name ruby 2.3.0dev (2015-10-10 trunk 52102) [x86_64-linux]
vm1_yield* 0.410

ruby 2.3.0dev (2015-10-11 trunk 52104)

total: 4 trial(s) (2 trial(s) for 2 benchmark(s))
....

benchmark results:
Execution time (sec)
name ruby 2.3.0dev (2015-10-11 trunk 52104) [x86_64-linux]
vm1_yield* 0.489

The difference is larger on http://rubybench.org/ruby/ruby/commits?result_type=vm1_yield&display_count=2000 so I thought I'll raise the issue.

Let me know if I can provide more information :)

[ko1 (Koichi Sasada)]

• Status changed from Open to Closed

Applied in changeset r53236.

---

• vm_insnhelper.c: move vm_callee_setup_block_arg() (and related
  functions) to the latter location.
  This moving recovers performance a little.
  [Bug #11829]

[usa (Usaku NAKAMURA)]

• Backport changed from 2.0.0: UNKNOWN, 2.1: UNKNOWN, 2.2: UNKNOWN to 2.0.0: DONTNEED, 2.1: DONTNEED, 2.2: DONTNEED